1

Chapter 1.6. Evolution While You Are Watching

Chapter 1.6 describes more or less direct observations of evolution. Although

Macroevolution which results in really profound changes of living beings occurs too

slowly for such observations, substantial changes in a lineage can and do occur at the

time scale of thousands or even hundreds of generations. Fast evolution is usually driven

by strong selection and can involve numerous allele replacements. Direct observations of

evolution come from data of four kinds.

Section 1.6.1 treats observations of rapid evolution of nature. Here we will

understand the concept of observation broadly, and include data on changes of

phenotypes in continuous paleontological records and on local adaptation. However,

under some circumstances evolution proceeds so rapidly that substantial changes occur in

the course of decades, and, thus, can be literally observed.

Section 1.6.2 considers domestication, a phenomenon which became common

after the origin agriculture ~12,000 years ago, and evolution of domesticated plants and

animals. Due to artificial selection and to natural selection under new environment,

important phenotypes of many domesticated species are currently way outside the range

of their natural variation, and the diversity of varieties is astonishing. Thus, domesticated

species provide a rich source of data on evolution at a relatively small scale.

Section 1.6.3 reviews data on experimental evolution in controlled populations of

a variety of organisms. Such experiments are an important tool of evolutionary biology,

which made it possible to directly observe a number of key phenomena. However, a

limitation of evolutionary experiments is that they must deal with organisms with short

generation times.

Section 1.6.4 presents data on rapid evolution of human pathogens, including

eukaryotes, bacteria, viruses, and cancerous cells. Dealing with the consequences of this

evolution leads is one of the key challenges to modern medicine. New pathogens emerge

regularly, sometimes with devastating consequences, and existing pathogens evolve

rapidly in order to evade the response of the immune system and to acquire resistance to

antibiotics. Evolution of malignant cells leads to progression of cancers.

Chapter 1.6 provides a transition from the data on past evolution, the subject of

Part 1, to Microevolution, treated in Part 2.

2

Section 1.6.1. Observations of evolution in nature

Rare instances of detailed, continuous paleontological records offer a precious

opportunity to effectively observe evolution at relatively long time scales. Differences

between tightly related geographical populations reveal their recent evolution, which

often can be explained by adaptation to local environments. Noticeable evolution of a

natural population can occur quite rapidly even in an undisturbed environment. There is a

wide variety of situations where rapid evolution was triggered by a recent anthropogenic

impact on the environment of a species.

1.6.2.1. Continuous fossil records

In a sense, a complete fossil record provides a videotape of past evolution, instead

of just a series of still pictures. Unfortunately, fossils mostly preserve information only

on external morphology, so that no records, except of very recent past, are complete in

the sense of providing data on evolution of genomes (Section 1.2.3). Still, even data on

external morphology alone can be very interesting. Microscopic organisms from marine

plankton is the only king of organisms which often leave rather detailed fossil records

(Section 1.2.3.3). However, under unusual circumstances such records are also possible

in other kinds of organisms. Let us consider two examples.

The first of them is evolution of fishes from genus Semionotus in Mesozoic rift-

valley lakes in North America. The North American and African plates were adjacent in

the Late Triassic and Early Jurassic, and tectonic activity which eventually opened the

Atlantic Ocean between them, also created deep rift valleys in what is now Eastern

United States. Due to quasy-periodic changes of the climate caused by Milankovich

cycling (Section 1.2.1.2), the basins of these valleys went through regular periods of lake

formation and regression, known as Van Houten cycles. Each Van Houten cycle lasted

~22,000 years and consisted of three phases: lake origin and expansion, existence of a

perennial lake up to ~100 m deep, and lake contraction and eventual disappearance.

During the second phase of a Van Houten cycle, sediments deposited in a lake consisted

of pairs of layer, each pair corresponding to a year (Fig. 1.6.2.1a). The longest continuous

3

succession of Van Houten cycles spans the period of 30 My (230-200 mya), and

corresponds to 4,400 m of sediments.

4

Fig. 1.6.2.1a. (top left) Basins of rift valleys in Eastern North America in Late Triassic

and Early Jurassic, where the lakes were located during phase 2 of each Van Houten

cycle. (top right) Deposits during a typical Van Houten cycle. (bottom left) An

excavation site in Newark Basin. (bottom right) Each year, a pair of layers, organic-rich

black and calcium carbonate-rich white, was deposited in the lake (McCune A. R.,

Diversity and Speciation of Semionotid Fishes in Mesozoic Rift Lakes, 2004).

Multiple species of an extinct genus of actinopterygian fishes Semionotus were

common in these Eastern North American rift-valley lakes (Fig. 1.6.2.1b). Semionotus

fossils from several Van Houten cycles in the Newark Basin are abundant enough to

document rapid speciation within this genus in the course of the first 5000-8000 years of

a cycle. During one particularly well-studied cycle, known as P4, from which fossils of

over 2000 Semionotus individuals have been found, at least 6, and perhaps as many as 13,

new species evolved within the Newark Basin rift-valley lake. The uncertainty is due to 7

species for which it is not clear whether they originated in situ or immigrated from other

lakes. The large number of individuals made it possible to even study morphological

variation within species (Fig. 1.6.2.1c). The dynamics of speciation in Mesozoic North

American rift-valley lakes is remarkably similar to that in modern African rift-valley

lakes (Chapter 1.1).

5

Fig. 1.6.2.1b. Fossils of Semionotus greenwoodi (top) and S. saginatus (bottom).

Fig. 1.6.2.1c. Speciation in Semionotus during phase 2 of one of Van Houten cycles (P4)

in the Newark Basin lake, recorded in almost 3 m of sediment. Each centimeter of

sediment corresponds to about nine years. (left) The number of species chronological

ranges that pass through a given layer of sediments. (middle) The first appearances of

new species, indicated by fish symbols. Fish outlines mark the first appearances in which

there is only one specimen, and filled outlines are species known from multiple

individuals. Arrows indicate the layers in which individuals with anomalies in

diagnostically important dorsal ridge scales were found. (right) The number of

individuals per centimeter of sediment, for which the dorsal ridge scales are visible.

The second example is a recent evolution of a gastropod genus Cerion on the

Great Inagua island (Bahamas). A very large number of shells can be collected there, and

their ages can be determined by the amino acid racemization and radiocarbon techniques

(Chapter 1.2). For most shells, racemization only racemization data were obtained, but

6

they were calibrated using radiocarbon clock. The data show that in the course of the last

~20,000 years the shape of the shell underwent substantial changes from smooth, tall, and

parallel-sided ancestral morphology to shorter, rounded, and rugged shells of extant

snails (Fig. 1.6.2.1d). Perhaps, this transition was caused by hybridization with an

invading species. These two examples demonstrate that a detailed enough fossil record

may reveal a continuous nature of evolution.

Fig. 1.6.2.1d. Changes of shell morphology in a species of genus Cerion on the southern

shore of the Great Inagua island in the course of the last ~20,000 years (Science 274,

1894, 1996).

1.6.2.2. Local adaptation

Any comparison of two extant forms of life indirectly reveals evolution of their

lineages after their divergence from the common ancestor. However, when the two forms

of life are very close to each other, so that their last common ancestor lived very recently,

such a comparison comes close to being a direct observation of evolution. Let us consider

several examples of this kind, all of which involve forms of life so similar to each other

that they can freely interbreed and are classified as different populations of the same

species. In this case, their conspicuous differences are regarded as local adaptations.

7

Often, local adaptations are observed when the range of a species spans different

elevations. A well-studied case is sticky cinquefoil Potentilla glandulosa, whose range in

California extends from the sea level to the peaks of Sierra Nevada (Fig. 1.6.2.2a). In

particular, moderate-altitude and high-altitude populations, which live under very

different environments, differ from each other in a number of morphological and

physiological traits. Transplant experiments show that plans from each population have a

much higher fitness in their native environment. Moreover, variable F2 offspring from

crosses between individuals from these two populations have higher fitness in the

environment of the parent to which an offspring is more similar morphologically.

8

Fig. 1.6.2.2a. Potentilla glandulosa from a low-altitude (left) and a high-altitude (right)

population (check!), its distribution in coastal California, and the scheme of an

experiment on its local adaptations (Evolutionary Ecology Research 7, 105, 2005).

Our species provides a number of cases of local adaptation. The most

conspicuous of them is between-population variation in the skin color (Fig. 1.6.2.2b),

which is due to variation of the amount of two pigments, pheomelanin (red) and

eumelanin (very dark brown). Genetically, polymorphisms at at least 3 loci with major

effects and probably to many more loci with minor effects are responsible for this

variation. Dark skin is obviously adaptive in the tropics, because it protects from

sunburns and skin cancers. It is less clear, however, why light skin, which is a derived

trait state, is apparently adaptive at high latitudes. Probably, dark skin prevents synthesis

of vitamin D when the amount of sunlight is limited, which can lead to rickets and

osteoporosis. One fact supporting this explanation is that Inuits, who live receive only a

very limited amount of sunlight but whose diet contains a lot of vitamin D, have

relatively dark skin. Other examples of local adaptation in humans include the ability of

adults to digest the milk sugar lactose, which independently evolved in Africa and

Northern Europe after the cow became domesticated (Section 1.5.3.1, Part 2), and

different adaptations to hypoxia which evolved independently in populations that live at

high altitudes in Andes, Tibet, and Ethiopia.

9

Fig. 1.6.2.2b. Geographical variation of skin color in Homo sapiens.

Often, local adaptations are due to differences in biotic, instead of abiotic,

components of the environments of different popualtions. For example, parasites are

often better adapted to attack hosts from the same locality (Fig. 1.6.2.2c). Parallel

geographical variation of Mullerian mimics is another example of this kind (Fig.

1.5.2.5c).

Fig. 1.6.2.2c. Parasitic trematode Microphallus sp. and its gastropod host Potamopyrgus

antipodarum (top). When parasites and hosts from two lakes, Mapourika and

10

Alexandrina, were studied, parasites from each lake infected hosts from the same lake

more efficiently (bottom).

1.6.2.3. Rapid evolution in undisturbed environments

Let us now consider observations of evolution which are truly direct, in the sense

that a researcher can directly observe an evolving lineage throught the course of

evolution. One may think that such observation must be impossible: if it took ~300,000

generations for the brain size to increase from 350cm3 to 1,400cm

3 in the human lineage

(Chapter 1.4), any changes in the course of a small number of generations that fit within

the career of a researched must be negligible. Fortunately, this is not always the case, due

to two reasons.

First, although the per generation rate of evolution of simple quantitave traits is,

indeed, usually very low, some other traits can evolve must faster. In particular, this is

true for traits that are responsible for interactions between species. One can say that even

a very long evolutionary trajectory can fit into a limited volume of a multidimensional

space of complex traits. Second, it may not be necessary for a researcher to observe

evolution as it occurs. Instead, evolution can be directly investigated from a record left in

deposits of some kind, as long as living organisms, instead of fossils, can be obtained

from this record. Deposits that can yield living organisms include Antrarctic and

Greenland ice cores (bacteria), Siberian permafrost (bacteria and plant seeds), and

sediments from lakes (bacteria and resting eggs).

In one study, dormant eggs of water flea Daphnia magna and its bacterial

endoparasite Pasteuria ramosa (Fig. 1.6.2.3a) were obtained from 24cm of sediments

deposited in a shallow pond in Belgium over the course of the last 40 years. Eggs were

hatched and exposed to parasite from sediments from the previous year, the same year,

and the following year. Daphnias were on average more susceptible to infection by

contemporary parasite (infectivity 0.65) than by parasite from the previous (infectivity

0.55) or the following (infectivity 0.57) year. This result is consistent with rapid

coevolution in this host-parasite system, a pattern known as Req Queen and predicted by

theory (Chapter 3.4). Briefly, changes the host keeps evolving to evade the parasite, and

the parasite keeps evolving to better infect its constantly changing host.

11

Fig. 1.6.2.3a. Daphnia magna, healthy (left) and infected with Pasteuria ramosa (Nature

450, 870, 2007).

Another example of extremely rapid evolution is also concerned with host-

parasite interactions. Many insects harbor bacteria from genus Wolbachia, which are

transmitted from mothers to offspring, and in some cases are lethal to male, but not too

female, offspring (Chapter 3.3). An insect population where infection by male-killing

Wolbachia is common is, thus, characterized by a female-biased sex ratio. In 2001, males

represented only 1% of individuals in populations of the butterfly Hypolimnas bolina on

Samoan islands Upolu and Savaii, and almost 50% of all eggs failed to hatch. However,

in 2006, after only about 40 generations of the butterfly, 1:1 sex ratio and high egg

vaibility were restored on Upolu and in one of the two studied populations on Savaii (Fig.

1.6.2.3b). This occurred due to evolution of resistance of the host to the male-killing

effect of the parasite, because the infection rate remained at 100%. Theory predicts that

natural selection for restoration of 1:1 sex ratio can be very strong (Chapter 3.4), which is

consistent with such rapid evolution. Analysis of sexes and Wolbachia infection in

butterflies from museum collections demonstrate that sex ratios and resistance to

Wolbachia kept evolving rapidly in various Pacific populations of H. bolina in the course

of the XX century.

12

Fig. 1.6.2.3b. Hypolimnas bolina female (top left) and male (top right), Wolbachia inside

the host cell (bottom left), and data of sex ratios in broods produced by individual

females and on total numbers of female and male offspring and median egg hatch rates in

the four populations studies (bottom right) (Science 317, 214, 2007; Current Biology 19,

1628, 2009).

1.6.1.4. Adaptation to recent changes of the environment

In addition to a number of well-documented cases of rapid, ongoing evolution in

apparently undisturbed habitats, there are a lot of data on such evolution which was

clearly caused by recent anthropogenic changes of the environment. Such changes can be

broadly subdivided into (i) changes of the pre-existing environment of a population, (ii)

creation of an essentially novel environment, and (iii) exposure of a species to a new

environment, after it invades a new area. Anthropogenic changes of each of these three

kinds offer many fascinating examples of rapid evolutionary responce.

13

A classical case of evolution triggered by changes in the pre-existing environment

is industrial melanism of peppered moth Biston betularia in Britain. In peppered moth,

the ancestral phenotype is light-gray (morph typica), which provides some protection

against birds on tree trunks that are covered by lichens. In the XIX century, lichens living

on tree truncs died in many areas of Britain, due to pollution of the air caused by coal

burning. As a result coloration of morpha typica was no longer protective, and morpha

carbonaria became common in the affected areas (Fig. 1.6.1.4a). The difference between

the two morphs is controlled by one polymorphic locus with two alleles, with morphs

typica and carbonaria corresponding to different homozygotes and the intermediately

colored morpha insularia corresponding to heterozygotes.

14

Fig. 1.6.1.4a. (top) Two extreme morphs of Biston betularia, typica (light) and carbonaria

(dark), on a tree trunk with (left) and without (right) lichens. (bottom) Dynamics of the

geographical distribution of the carbonaria morhp: (a) Mid-XX century; (b) 1983-1984;

(c) 1987-1999 (Quarterly Review of Biology 78, 399, 2003).

At mid-XX century a steep cline of carbonaria frequency running from the north

of Wales to the southern coast of England separated a region of 90% frequency to northeast. At that time pollution was greatly reduced. By the

1980s the plateau of 90% frequency of carbonaria had contracted to northern England.

The frequency of carbonaria has since continued to drop so that the maximum is now less

than 50% and in most places below 10% (Fig. 1.6.1.4a). There have been similar declines

in Europe and North America.

Both the past origin and the ongoing disappearance of industrial melanism in B.

betularia are so rapid that strong selection must be involved. The observed rate of

reduction of carbonaria frequency requires 5%-20% selection against it. Experiments that

involved predation by birds show a net advantage to carbonaria in regions where typica

frequencies were low at the time of the experiment, and a disadvantage where typica

frequencies were high. This would be expected if environment and frequency were

associated, and selective predation played a part in generating the association.

Another important form of pollution is pollution of soil by metals, such as zinc,

copper, and lead, due to mining and other industrial activities. Many plant species rapidly

adapted to this pollution, and evolved the ability to thrive in the presence of high

concentrations of metals that are lethal to ancestral genotypes. For example, rockcress

15

Arabidopsis halleri now consists of a number of metallophyte (M) populations that can

grow in the presence of heavy metals, as well as ancestral of nonmetallophyte (NM)

populations that cannot (Fig. 1.6.1.4b). Phylogenetic analysis demonstrated that different

M populations are not closely related to each other and, instead, that each of them

originated locally from the ancestral NM populations. are more genetically related to

their closest NM populations than to each other. Thus, evolution towards increased metal

tolerance that created distinct M population groups represents one more example of

homoplasy. In many plant species individual from M populations lose in competition to

individuals from NM populations others in pollution-free habitats, indicating that metal

tolerance is involved with a substantial cost.

16

Fig. 1.6.1.4b. (top) Rockcress Arabidopsis halleri. (middle) Geographic distribution of A.

halleri genotypes, with M populations shown by red dots and NM populations shown by

blue dots. (bottom) Phylogeny of A. halleri populations, showing close relatedness

between M and NM populations from adjacent localities (Molecular Ecology 14, 4403,

2005; Genetica 112, 165, 2001).

Evolution of resistance to pesticides is another common result of chemical

pollution of the environment. Pesticides, such as DDT (dichlorodiphenyltrichloroethane)

may cause severe selection in insects, both their intended targets and others, and this

selection often leads to rapid response, due to allele replacements at one ore few loci.

Evolution of various degrees of resistance to one or several pesticides has been

documented in hundreds of species of arthropods.

Rapid evolution of wild populations can also be caused by their intensive and

selective harvesting or some period of time. Selective harvesting of large individuals

often leads to evolution of early maturation and reduced growth rates, among other trait

states. One can view selective harvesting as a form of artificial selection imposed on a

wild population. In one striking case, rapid evolution of earlier maturation and small size

17

preceded the collapse of the population of Atlantic cod Gadus morhua off southern

Labrador and eastern Newfoundland, caused by overfishing. Fishing for this cod

populations was banned in 1992, but in more than a decade afterwards these life history

traits did not revert to their original states and the population size remained historically

low (Fig. 1.6.1.4c).

Fig. 1.6.1.4c. Atlantic cod Gadus morhua and its temporal trends in the age at 50%

maturity (a), annual length increments (b) and annual survival probabilities (c) of cod

from 3 locations (solid, dotted, and dashed lines) off southern Labrador and eastern

Newfoundland (Nature 428, 932, 2004; Nature 457, 803, 2009).

Rapid evolution of a species can also happen after it colonizes an artificial

environment. The ancestral coloration of rock pigeons Columbia livia include pale grey

wings with two black bars. However, a large proportion of pigeons living in the urban

environment either have uniformly dark plumage (melanists) or checkered plumage with

dark and grey spots (Fig. 1.6.1.4d). This variation is due to at least 2 major genes and

some minor genes. By the end of World War II, urban populations of pigeons in major

18

cities of the western part of the Soviet Union were mostly extinct, and were later

reestablished form grey-colored pigeons which immigrated from the rural habitats.

However, urban melanism evolved soon after this, apparently independently in in

different cities. It is not yet clear why melanism is advantageous under urban

environments, but selection favoring it must be quite strong.

Fig. 1.6.1.4d. Pigeons of various colorations, including ancestral light grey, melanistic,

and checkered, in a city square.

Recent invasions, almost universally caused by human activity (Section 1.3.3,

Section 1.5.1.6) led to a variety of evolutionary phenomena. Invasion of a species into a

new range exposes it to novel conditions and often trigger rapid adaptive evolution. For

example, cane toads Bufo marinus were introduced to Australia from Hawaii in 1935 to

control insect pests in sugar-cane fields. Since then, the range of these toxic and highly

invasive toads in Australia expanded dramatically, and the rate of this expansion is now

about 5 times higher than initially (Fig. 1.6.1.4e). This acceleration is due to evolution of

higher endurance in toads from the invasion front populations, which is apparently

favored by an opportunity to colonize new areas which are not yet occupied by toads.

19

Fig. 1.6.1.4e. Cane toad Bufo marinus and its current range in Australia (Nature 439,

803, 2006; Oecologia 16, 343, 2010).

Adaptive evolution after an invasion can result in formation of host races in a

phytophagous insect, that can eventually lead to speciation (Chapter 2.6). On example of

this kind is provided by an aphid Acyrthosiphon pisum, which evolved into two host

races after its invasion into North America (Fig. 1.5.2.4e). However, host races may also

result from an invasion of a new host plant. Despite its name, apple maggot Rhagoletis

pomonella had only one native host in North America, and this host was species of

hawthorn, Crataegus spp (Fig. 1.6.1.4f). However, a new host race of Rh. pomonella

evolved in the middle XIX century, on introduced domesticated apple Malus pumila. The

hawthorn and apple host races of Rh. pomonella experience substantially different

selection, and have already evolved a substantial degree of reproductive isolation, thus

representing a pair of incipient species.

20

Fig. 1.6.1.4f. Rhagoletis pomonella on its original host, a hawthorne (left) and on its

introduced host, an apple tree (right) (Nature 407, 739, 2000).

Often, an invasion leads to absolutely unexpected consequences. A spectacular

example of such a situation is provided by introduction of the P-element into the gene

pool of Drosophila melanogaster. P element is a TE which was originally restricted to

several species New World species of Drosophila. D. melanogaster, as well as other Old

World species of Drosophila, never had P element-like TEs. After its introduction into

Americas, perhaps in the XIX or even in the XX century, D. melanogaster became

infected with P element. The initial infection was probably due to a unique horizontal

transmission event from D. willistoni, who has P elements nearly idential to that now

present in D. melanogaster, and may be mediated by a parasitic mite (Fig. 1.6.1.4g). This

transmision was followed by an extremely rapid expansion of the P element within D.

melanogaster. In early XX century, all wild-caught D. melanogaster were P-negative,

and old lobaratory strains derived from those flies remain P-negative. However, by the

end of the XX century every D. melanogaster in the wild throughout its global range,

including the Old World, was P-positive. In fact, P element is deleterious to the host, but

enjoys a strong transmission advantage, because it takes just one infected parent to infect

all the offspring. Thus, a single selfish TE took over the whole species with a global

range in under 100 years.

21

Fig. 1.6.1.4g. Males of Drosophila willistoni (left), and D. melanogaster (right), and

distribution the current distribution of P elemenent within some Drosophila species

(Bioessays 14, 681, 1992; Chromosoma 110, 148, 2001).

Section 1.6.2. Evolution of domesticated animals and plants

Many species of animals and plants were domesticated beginning from 12,000

years ago, and wild progenitors of most of them still exist. A domesticated species

breeds in captivity, under the influence of both natural and artificial selection. As a result,

domesticated species, including some of those domesticated recently, evolve phenotypes

that are very different from phenotypes of their progenitors and split into many

amazingly diverse varieties. Artificial selection can be particularly efficient when only

22

one or a small number of traits is targeted. However, evolution of domesticated plants

and animals is also involved with a substantial cost, wich emerges due to several

genetical processes.

1.6.2.1 Phenomenon of domestication

Each domesticated animal or plant species evolved from a wild progenitor or, in

some cases, from several tightly reated progenitors. Direct evidence of ancient

domestications come from paleontological and archeological data. For example, the

earliest fossils of a dog, found in Europe, are 30,000 old, making dog the likely first

domesticated mammal. Among birds, the first domesticated species was probably rock

pigeon, and Egyptian hieroglyphics and Mesopotamian cuneiform tablets mention

domesticated pigeons over 5,000 years ago. Numerous evidence show that the first

domesticated livestock species were sheep, goat, pig, and cow, and that barley, two

species of wheat, rice, and corn were among the first domesticated plants.

Crucial information on ancient domestications in also provided by comparative

analysis of modern breeds of a domesticated species and of extant populations of its

progenitor. Indeed, with the notable exception of aurochs and wild horses, which recently

became extinct in the wild, progenitors of all other common domesticated species are still

around. Thus, a wild progenitors and diverse varieties of a domesticated species which

arose from it together provide a unique opportunity to simultaneously study an essentially

unmodified ancestor and its diverse descendants. Indeed, the time since even the earliest

domestications is trivial at the scale of rates of typical rates of evolution, leaving a wild

progenitor with only minimal changes. However, adaptation of its domesticated

descendants to a drastically altered enviroment, together with artificial selection, always

causes very substantial changes. Historically, some domesticated species were given

names different from names of their progenitors. Table 1.6.2.1a summarizes currently

available data on domestication of some of the most important species. It is likely that

future will bring some revisions.

Table 1.6.2.1a. Data on ancient domestication of some important species

Domesticated Time of Number of Region(s) Subsequent Reference

23

species and its

wild progenitor

domestication

(years BP)

domestication

events

where

domesticati

on occurred

extensive

hybridization

with the wild

progenitor

dog Canis lupus;

grey wolf C.

lupus

>31,000 1 (?) Near East yes, in East

Asia and

Europe

Nature 464, 898, 2010

cat Felis catus;

wildcat Felis

sylvestris

> 9,5000 1 Near East limited Science 317, 519, 2007

goat Capra

hircus; bezoar

Capra aegagrus

11,000 2 Near East no? PNAS 105, 17659, 2008

sheep Ovis

aries; mouflon

Ovis orientalis

11,000 3 Near East limited Science 324, 532, 2009

pig Sus scrofa,

wild boar Sus

scrofa

10,500 several Near East,

East Asia

and Europe

limited Molecular Phylogenetics and Evolution 54, 680, 2010

cow Bos taurus,

aurochs Bos

primigenius

10,000 several Near East,

North

Africa

limited PLOS ONE 4, e5753, 2009

horse Equus

cabalus; wild

horse Equus

ferus

6,000 several Central

Asia

yes Science 326, 865, 2009

pigeon Columba

livia; rock

pigeon

Columbia livia

~10,000 1 Near East,

Egypt

limited Biological Journal of the Linnean Society 93, 657, 2008

chicken Gallus >8,000 at least 2 South-East no? PLoS Genetics 4, e1000010,

24

gallus; red

junglefowl

Gallus gallus

and grey

junglefowl

Gallus

sonneratii

Asia 2008

Duck Anas

platyrhynchos;

mallard Anas

platyrhynchos

2500 1 (?) China no? BMC Evolutionary Biology 8, 298, 2008

Goldfish

Carassius

gibelio; gibel

carp Carassius

gibelio

1700 1 (?) China no Gene 430, 5, 2009

Barley Hordeum

vulgare; wild

barley Hordeum

spontaneum

10,000 2 Near East no? PNAS, 104

3289, 2007

Emmer wheat

Triticum

dicoccum; wild

emmer Triticum

dicoccoides

9,500 1 Near East no? Israel Journal

of Plant

Sciences 55,

207, 2007

Einkorn wheat

Triticum

monococcum;

wild wheat

Triticum

boeoticum

9,000 1 Near East no? Genome 51,

444, 2008

25

Rice Oryza

sativa, wild rice

Oryza rufipogon

10,000 2 South Asia

and China

no? PNAS 103,

9578, 2006

Corn Zea mays

mays; teosinte Z.

mays ssp.

parviglumis and

mexicana

9,000 1 Mexico,

Balsas

valley

yes, corn

originated from

Z. m.

parviglumis,

but genes from

Z. m. mexicana

were admixed

later

PNAS 106, 4957, 2009

Potato Solanum

tuberosum; wild

potato Solanum

brevicaule

5,000 1 Northern

Peru

limited PNAS 102, 14694, 2005

Tomato

Solanum

lycopersicum;

wild tomato

Solanum

lycopersicum.

3,000 (?) 1 (?) Andean

Region

yes Annals of

Botany 100,

1085, 2007

With the exception of dog, cat, and pigeon, all other ancient domestications were

directly related to agriculture. In fact, domestication of several species of crop grasses

and of livestock, which still serve as the staple foods for humanity, heralded the

beginning of agriculture at the onset of the Holocene interglacial ~11,700 years ago (Fig.

1.2.4.1b; Fig. 1.3.3.3a). Transition from foraging to agriculture was likely triggered by

the end of the last glacial period and, more precisely, of its last cold segment between

12,800 and 11,500 years ago, known as Younger Dryas stadial. Extinction of large game

mammals, due to hunting, was another likely factor in this transition (Section 1.3.3).

Agriculture originated independently and almost simultaneously in at least four

regions: the Fertile Crescent in the Middle East, the Yangtze reion of Southeast Asia,

26

Mesoamerica, and the Andean region in South America, followed by its mostly

independent development in several other regions. Many of these regions are small

enough to be regarded as centers of plant domestication (Fig. 1.6.2.1a). Fig. 1.6.2.1b

presents some details about the Fertile Crescent center.

Fig. 1.6.2.1a. Centres of plant domestication. Solid-shaded areas and hatched areas

indicate regions of important seed-crop domestication and vegecultural crops,

respectively. Accepted primary domestication centres are shown in black, and potentially

important secondary domestication centres are shown in grey. Arrows indicate major

trajectories of spread of agriculture and crops out of some centres. Areas are numbered,

and examples of crop species and the year by which they were domesticated in each area

are as follows: 1, eastern North America (Chenopodium berlandieri, Iva annua and

Helianthus annuus, 4,5004,000 years before present (yr bp)); 2, Mesoamerica

(Cucurbita pepo, 10,000 yr bp; Zea mays, 9,0007,000 yr bp); 2a, northern lowland

neotropics (Cucurbita moschata, Ipomoea batatas, Phaseolus vulgaris, tree crops, 9,000

8,000 yr bp); 3, central mid-altitude Andes (Chenopodium quinoa, Amaranthus caudatus,

5,000 yr bp); 3a, north and central Andes, mid-altitude and high altitude (Solanum

tuberosum, Oxalis tuberosa, Chenopodium pallidicaule, 8,000 yr bp); 3b, lowland

southern Amazonia (Manihot esculenta, Arachis hypogaea, 8,000 yr bp); 3c, Ecuador and

northwest Peru (Phaseolus lunatus, Canavalia plagiosperma, Cucurbita ecuadorensis,

10,000 yr bp; the question mark indicates that there is some question of the independence

of crop origins of this centre from 3, 3a and 3b); 4, West African sub-Sahara (Pennisetum

27

glaucum, 4,500 yr bp); 4a, West African savanna and woodlands (Vigna unguiculata,

3,700 yr bp; Digitaria exilis, Oryza glaberrima, 4,000 yr bp?); 6, East African uplands (Eragrostis tef,

Eleusine coracana, 4,000 yr bp?) and lowland vegeculture (Dioscorea cayenensis, Ensete

ventricosum, poorly documented); 7, Near East (Hordeum vulgare, Triticum spp., Lens

culinaris, Pisum sativum, Cicer arietinum, Vicia faba, 13,00010,000 yr bp); 7a, eastern

fertile crescent (additional Hordeum vulgare, 9,000 yr bp, and also goats); 8a, Gujarat,

India (Panicum sumatrense, Vigna mungo, 5,000 yr bp?); 8b, Upper Indus (Panicum

sumatrense, Vigna radiata, Vigna aconitifolia, 5,000 yr bp); 8c, Ganges (Oryza sativa

subsp. indica, 8,5004,500 yr bp); 8d, southern India (Brachiaria ramosa, Vigna radiata,

Macrotyloma uniflorum, 5,0004,000 yr bp); 9, eastern Himalayas and Yunnan uplands

(Fagopyrum esculentum, 5,000 yr bp?); 10, northern China (Setaria italica, Panicum

miliaceum, 8,000 yr bp; Glycine max, 4,500 yr bp?); 11, southern Hokkaido, Japan

(Echinochloa crusgalli, 4,500 yr bp); 12, Yangtze, China (Oryza sativa subsp. japonica,

9,0006,000 yr bp); 12a, southern China (Colocasia, Coix lachryma-jobi, poorly

documented, 4,500 yr bp?); 13, New Guinea and Wallacea (Colocasia esculenta,

Dioscorea esculenta, Musa acuminata, 7,000 yr bp) (Nature 457, 843-849).

28

Fig. 1.6.2.1b. Domestication in the Fertile Crescent center. Shaded areas indicate the

approximate areas of domestication of pig, cattle, sheep, and goats with the likely dates

of initial domestication. Colored lines enclose the wild ranges of Einkorn wheat, emmer

wheat, and barley. Green-shaded area in southern Levant indicates the region where all 3

grains were first domesticated 12,000 years ago (PNAS 106 Supplement 1 9971, 2009).

With few exceptions, domesticated species are either flowering plants or

mammals. Still, even within these clades only a small fraction of species became

domesticated. Among ~200,000 species of flowering plants, only several hundreds have

been truly domesticated. Among 148 species of mammals weighting over 45 kg, only 14

were domesticated, and domestication of the remaining ~5,000 species produced

proportionally even less domesticates. Strikingly, the most important domesticated

species (Table 1.6.2.1a) were also the first to be domesticated, and recent domesticates

are of only secondary significance. Clearly, the choice of species that became

domesticated was non-random and restricted.

To become domesticated, a plant species must be useful, first of all for food.

Other traits are much less important, as a majority of plants can be successfully grown

after being planted artificially, and even a relatively long generation time was not an

obstacle for domestication of a number of trees. In contrast, the conditions required for

domestication of an animal species are more restrictive, due to complex behavior of

animals. In addition to being useful, a wild progenitor animal must be amenable to

taming, have a diet that can be realistically provided, and be capable of breeding in

captivity. Thus, zebras, anteaters, or pandas must inherently difficult to domesticate.

Still, at least some wild mammals with aggressive and suspicious behavior can be

domesticated, due to evolution under artificial selection. This has been demonstrated by

experimental domestication of the silver fox, a color variation of the wild red fox Vulpes

vulpes. The experiment begun in 1959 at the Institute of Cytology and Genetics in

Novosibirsk in the laboratory of Dmitry Belyaev, and involved artificial selection of

captive foxes for only one trait, tame and friendly behavior. A substantial progress was

made in less than 20 generations and continued afterwards (Fig. 1.6.1.2c). Tame foxes

exhibit highly social behavior with both other members of their own species and humans

29

in a playful, friendly manner. In contrast, foxes from an unselected population, as well as

from a strain bred for aggressive behavior, avoid social interactions with humans.

Fig. 1.6.1.2c. Wild foxes (top) and domesticated foxes after 40 generations of selection

for tame behavior. (Journal of Heredity 95, 185, 2004).

A host of additional changes never deliberately selected for also appeared in

domesticated foxes. They included piebald coat color, drooping ears, shorter,

occasionally upturned tails, shortened snouts and shifts in the developmental timing of

various other characteristics. As they aged, many of the fox pups began behaving in a

manner similar to domestic dogs by barking, whining and licking their handlers. Thus,

even when substantial changes are required for domestication, artificial selection can

produce them in a remarkably short time.

Ancient domestications took a much longer time. Archeological data indicate that

in plants prolonged precultivation harvesting of a wild progenitor often preceded its

cultivation, and the onset of cultivation was followed to slow emergence of domestication

traits (for example, Fig. 1.6.2.1d). Ancient domestication of a animal must also be a slow

30

process. Consistent application of strong artificial selection and other techniques which

can cause a rapid evolution of a captive population became common only in more recent

times.

Fig. 1.6.2.1d. A summary of the origins of agriculture in the Fertile Crescent and the

subsequent development of wheat and barley cultivation worldwide. A series of

continuous human activities have led to an equally continuous series of changes to crop

plants. The human activities began with a variety of human interventions in the growth of

wild plants. These interventions prompted the gradual emergence of domestication traits

in the crop plants, the first indications of these traits being seen in the archaeobotanical

record around 12,500 years BP. Eventual fixation of these traits by 9000 BP enabled

agriculture to spread beyond its area of origin. The spread of agriculture exposed crops to

new environments to which they adapted by further evolution. The nutritional and

culinary properties of locally adapted landraces also underwent change as agriculture

gradually intensified, ancient DNA analysis suggesting that glutenin alleles associated

with good breadmaking were present in wheat being grown during the Greek Bronze Age

at 3000 BP. The series of human activities continues today with the application of

directed breeding programmes which have resulted in the elite varieties of cereals that

support our modern society (TREE, 24, 103, 2008).

1.6.2.2. Evolution of new phenotypes in domesticated species

31

From the evolutionary perspective, the key distinction of a domesticated species

is that it breeds in captivity, and thus evolves separately from the progenitor that remains

in the wild. After domestication, evolution proceeds under joint action of artificial

selection and natural selection, which may be very different, due to a new environment,

from natural selection in the wild. As a result, domestication is always followed by rapid

evolution. As it is the case with evolution in the wild, evolution of a domesticated species

involves both anagenesis and cladogenesis, to be considered in this and the next Section.

The nature of selection may be very different in different domesticated species. In

the early history of domesticated plants, deliberate artificial selection was rare. Instead,

selection arose mostly as a result of cultivation practices, which favored survivial and

reproduction of plants with only some phenotypes. In contrast, artificial selection was

likely always applied deliberately to domesticated animals. In particular, postzygotic

selection, due to selective culling of animals, must be near-universal. In contrast,

prezygotic artificial selection, due to selective breeding, apparently became common only

much more recently. Obviously, joint action of postzygotic and prezygotic artificial

selection can be especieally efficient.

Comparison of evolutionary histories of different domesticated species reveal

extensive homoplasy, sometimes referred to as domestication syndrome. In cereal crop

plants, domestication syndrome involves seven components: loss of seed dispersal, due to

the evolution of tough rachis which results in the grain remaining attached to the mature

ear; loss of grain dispersal aids, such as hairs, hooks and awns; increase in grain size; loss

of sensitivity to environmental cues for germination and flowering; synchronous tillering

and ripening; compact growth habit; and enhanced culinary chemistry. Similar changes

occurred repeatedly in non-cereal crops, with additional homoplasious loss of defensive

armor and toxins, absent in even wild grasses. Domestication syndrome in mammals

consists of changes affecting a variety of traits (Fig. 1.6.2.2a). The For domestication

experiment suggests that many of these changes may be by-products of artificial selection

for tame behavior.

32

Fig. 1.6.2.2a. Domestication syndrome in mammals (J. of Zoology 269, 261, 2006).

Genetic changes behind early phenotypic evolution under domestication often

involve allele replacements with major effects. Such replacements are possible when

strong selection, natural or artificial, favors a substantially novel phenotype, as long as

mutations that move a phenotype much closer to the new optimum in a single leap are

available. As Darwin put it in a book "The Variation of Animals and Plants Under

Domestication", published in 1868: "Although man does not cause variability and cannot

even prevent it, he can select, preserve, and accumulate the variations given to him by the

hand of nature almost in any way that he chooses; and thus he can certainly produce a

great result". Spectacular examples of large-effect mutations and their fixations are

provided by domesticated corn and tomato.

Corn Z. m. mays is drastically different from its original wild progenitor teosinte

Z. m. parviglumis in the overall structure of the plant, as well as in the morphology of

female inflorescence and of kernels (Fig. 1.6.1.2b). Teosinte has long lateral branches

terminated by male inflorescences, and its female inflorescence, rachis, consists of only

two interwinded rows of kernels, each covered by a hard coating. In contrast, corn has

short lateral branches terminated by female inflorescences, ears, and each ear consists of

many rows of coatless kernels. Comparison between genomes of corn and its wild

33

relatives demonstrated that the corn-teosinte difference in the structure of lateral branches

is due to an allele replacement which occurred in non-coding DNA 60 kb upstream of the

protein-coding gene named tb1 (teosinte branched1). Apparently, this replacement

affectes regulation of transcription of tb1. The loss of hard kernel coating in corn is also

due to a single allele replacement, which occurred within the coding region of a gene that

encodes transcription factor tga1 (teosinte glume architecture1), and led to replacement

of a single amino acid of the protein. The difference between teosinte rachis and corn ear

is due to allele replacements at several loci, including tb1, which affects several traits at

once. Overall, a large number of allele replacements, with effects varying from drastic to

small, were responsible for evolution of domesticated corn.

Fig. 1.6.1.2b. Annual teosinte and the product of its domestication, maize. A substantial

portion of the phenotypic differences between maize and teosinte is explained by

variation in an enhancer of gene expression at the tb1 (teosinte branched 1) locus.

Intergenic sequences 58-69 kb 5' to the tb1 coding region confer pleiotropic effects on

maize morphology (Nature Genetics 38, 594, 2006; PNAS 106, suppl. 1, 9979, 2009).

A salient difference between wild and domesticated tomato is its fruit size (Fig.

1.6.1.2c). Analysis of offspring from crosses between the two (Section 2.2.3) revealed six

major-effect loci where allele replacements responible for this difference took place. In

particular, allele replacement(s) at locus fw2.2 (fruit weight 2.2) increases the fruit weight

by up to 30%, and may be the first one fixed in the course of tomato domestication. This

gene encodes a repressor of cell division, and its allele that increases the fruit size has an

34

altered promoter region, resulting in its reduced expression during the cell-division phase

of fruit development. Allele replacements at many other genes, with effects of different

magnitudes, are responsible for other new phenotypes in domesticated tomatoes,

including a wide variety of fruit shapes (Fig. 1.6.1.2c).

Fig. 1.6.1.2c. Variation in size and shape of tomato fruit. (A) The large-fruited tomato

Giant Heirloom common to modern agriculture (right) and the typical fruit of a related

wild species (Lycopersicum pimpinellifolium). (B) The range of fruit size and shape

variation in tomato. (C) Cross-section of the fruit from a plant homozygous for a

mutation at the fasciated locus causing multiple locules. (D) Alternate allele of fasciated

associated with unfused carpels. (E) Fruit from Long John, which carries mutations at

both the sun and ovate loci causing extremely long, narrow fruit. (F) Bell pepper-type

fruit produced by Yellow Stuffer. (G) Fruit from two different cultivars homozygous

for a mutation at the ovate locus. In the variety on the left, the ovate mutation results in

the production of fruit that are both elongated and constricted at the stem end of the fruit

(hence, the pear shape). However, in the processing variety on the right, the ovate

35

mutation causes elongated fruit but has a much reduced effect on neck constriction (The

Plant Cell 16, S181, 2004; Annals of Botany 100, 1085, 2007).

Of course, evolution of a domesticated species does not stop after the first phase

of domestication is over, but remains an ongoing process. In recent times, this evolution

actually accelerated, due to previously unavailable methods that produce variation,

through artificial mutagenesis and artificial lateral gene transfer (so that the relevant part

of the Darwin's quotation earlier in this section is no longer valid) and, even more

importantly, to more efficient artificial selection. For example, genome-level comparison

of the genotypes of eight modern breeds of chicken with the genotype of red junglefowl

(Fig. 1.6.1.2d) showed that an allele replacement at the locus for thyroid stimulating

hormone receptor (TSHR), which plays a key role in metabolic regulation and

photoperiod control of reproduction in vertebrates, occurred early in the course of

domestication, together with many other allele replacements. However, a number of

allele replacements, probably at loci controlling growth, appetite and metabolic

regulation, also occurred in broiler (meat-producing) breeds, after their divergence from

layer (egg-producing) breeds just a century ago.

36

Fig. 1.6.1.2d. Schematic tree based on breeding history (branch lengths not to scale) of

the chicken breeds used for comparative genome-level analysis. The minimum time of

divergence, when known, is indicated (Nature 464, 587, 2010).

When extreme states of a trait are favored by efficient artificial selection, it can

produce phenotypes far outside the range of natural variation. A red junglefowl hen lays

4-7 eggs per ear, and a leghorn hen can lay up to 280 eggs per year. Yields of modern

varieties of domesticated cereal grasses exceed that of their wild progenitors by more

than an order of magnitude. It seems that productivity of some domestic varieties may be

approaching the physiological limits.

1.6.2.3. Evolution of diversity within domesticated species

A domesticated species always becomes split into a number of lineages which

may occassionally hybridize with each other, as well as with the wild progenitor, but

mostly evolve independently under different modes of natural and artificial selection.

With time, this pattern results in an astonishing diversity of phenotypes in different

lineages (landraces, strains, and breeds) within a domesticated species. Fig. 1.6.1.2c and

Fig. 1.6.1.2d gave some idea of this diversity. Another classical example is the world's

oldest domesticated bird is rock pigeon Columbia livia. (Fig. 1.6.2.2a), whose natural

range includes western and southern Europe, North Africa, and southwest Asia. Fig.

1.6.2.2b presents a small sample of domestic breeds of the rock pigeon.

37

Fig. 1.6.2.2a The wild progenitor of domestic pigeon breeds, the rock pigeon.

Darwin was an expert in pigeons and was deeply impressed by their evolution in

captivity ("Origin of Species", Chapter 1):

Believing that it is always best to study some special group, I have, after

deliberation, taken up domestic pigeons. I have kept every breed which I could

purchase or obtain, and have been most kindly favored with skins from several

quarters of the world ... . I have associated with several eminent fanciers, and have

been permitted to join two of the London Pigeon Clubs. The diversity of the breeds

is something astonishing. Compare the English carrier and the short-faced tumbler,

and see the wonderful difference in their beaks, entailing corresponding differences

in their skulls. ... In the skeletons of the several breeds, the development of the

bones of the face, in length and breadth and curvature, differs enormously. The

shape, as well as the breadth and length of the ramus of the lower jaw, varies in a

highly remarkable manner.

Altogether at least a score of pigeons might be chosen, which, if shown to an

ornithologist, and he were told that they were wild birds, would certainly be ranked

by him as well-defined species. Moreover, I do not believe that any ornithologist

would in this case place the English carrier, the short-faced tumbler, the runt, the

barb, pouter, and fantail in the same genus; more especially as in each of these

breeds several truly-inherited sub-breeds, or species, as he would call them, could

be shown him. Great as are the differences between the breeds of the pigeon, I am

fully convinced that the common opinion of naturalists is correct, namely, that all

are descended from the Rock Pigeon (Columba livia), including under this term

several geographical races or sub-species, which differ from each other in the most

trifling respects.

38

Fig. 1.6.2.2b. Some domestic pigeon breeds.

Fod many domestication species comparison between genotypes of its breeds

revealed their phylogeny. Fig. 1.6.2.2c shows the phylogeny of dog breeds. Evolution of

domestic dog after its origin in the Near East involved interbreeding with local wolfs in

East Asia and Europe. Several breeds branched off rather early in the dog history,

although a majority of contemporary breeds are of rather recent origin. Among other

39

species that were domesticated a long time ago and where the phylogeny of breeds have

been reconstructed are cat (Fig. 1.6.2.2d), sheep (Fig. 1.6.2.2e), and corn (Fig. 1.6.2.2f).

40

Fig. 1.6.2.2c. Phylogeny of dog breeds, after their origin from grey wolves (Nature 464,

898, 2010), toghther with pictures of the oldest extant breeds: Basenji, Akuta, Chow

Chow, Dingo, Shar Pei, Alaskan Malamute, Siberian Husky, Afghan Hound, and Saluki.

41

Fig. 1.6.2.2d. Phylogeny of some breeds of cats (Science 317, 519, 2007; Genomics 91,

12, 2008).

42

Fig. 1.6.2.2e. The history of domesticated sheep involved two waves of migration.

Primitive breeds of sheep, which are remnants of the first migrations, are shown in the

figure. These breeds possess morphological characteristics (such as darker, coarser

fleece; moulting coat; frequent presence of horns in females) similar to those of the

mouflon. (A) Urial sheep; (B) Cyprus Mouflon; (C) Mediterranean Mouflon; (D) Orkney

sheep; (E) Soay sheep; (F) Gute sheep; (G) land sheep; (H) Icelandic sheep; and (I)

Hebridean sheep (Science 324, 532, 2009).

43

44

Fig. 1.6.2.2f. (top) Phylogeny of corn and teosinte. Dashed gray line circumscribes the

monophyletic domestic corn lineage. Asterisks identify those populations of ssp.

parviglumis basal to corn, all of which are from the central Balsas River drainage,

suggesting that this regions was the cradle of corn domestication. (a) Individual plant tree

based on 193 corn and 71 teosinte. (b) Tree based on 95 ecogeographically defined

groups. The arrow indicates the position of Oaxacan highland corn that is basal to all of

the other maize. (bottom) Geographic distribution of maize and teosinte used for the

phylogenetic analysis of maize domestication. Inset shows the distribution of the

34 populations of ssp. parviglumis in southern Mexico with the populations that are basal

to maize. The blue line is the Balsas River and its major tributaries. (PNAS 99, 6080,

2002; Ann. Rev. of Genetics 38, 37-59, 2004).

Perhaps, it is not surprising that a very substantial diversity of breeds can evolve,

under artificial selection, in the course of ~10,000 years. Indeed, hundreds of species of

cichlids evolved from one or a small number of ancestors in Lake Victoria during a

similar period of time ~12,000 (Chapter 1.1?). Still, most of dog and sheep breeds are, in

fact, much younger, and even species which were domesticated much later still evolved

and astonishing variety of breeds. Let us consider several examples of such species.

Goldfish is the result of domestication of a Chinese population of gibel carp

Carassius gibelio (Fig. 1.6.2.2g), which was kept there in captivity as edible fish. Wild-

type gibel carps has drab olive-grey body color. The red coloration was first recorded

during the Tsin dynasty (265419 CE). During the Tang Dynasty (618 - 907), it was

popular to raise gibel carps in ponds. As the result of a dominant mutation, one of these

fishes displayed "gold" (actually yellowish orange) coloration. People began to breed the

gold variety and began to display them in small containers. In 1162, the empress of the

Song Dynasty (960-1279) ordered the building of a pond to collect the red and gold

variety. As time passed, more mutations occurred, producing new color variations, and

fancier varieties of goldfish were developed. The occurrence of other colors was first

recorded in 1276. The first occurrence of fancy tailed goldfish was recorded in the Ming

dynasty (1368 - 1644). In 1502, goldfish were introduced to Japan, where the Ryukin and

Tosakin varieties were developed. In 1611, goldfish were introduced to Portugal and

from there to other parts of Europe.

45

Obviously, strong artificial selection was applied to coloration and external

morphology of domestic goldfish, usually in the direction opposite to that of natural

selection. Genetic variation on which artificial selection operated was partially present in

the wild progenitor population and partially provided by new mutations that occurred in

captivity. Some of the breeds of goldfish are shown in Fig. 1.6.2.2h, and the phylogeny of

several breeds, reconstructed on the basis of their mitochondrial DNA, is shown in Fig.

1.6.2.2i. Obviously, none of these fishes would do well in the wild.

Fig. 1.6.2.2g. Gibel carp Carassius gibelio, the wild progenitor of goldfish (Gene 430, 5,

2009).

46

Fig. 1.6.2.2h. A sample of breeds of domestic goldfish.

47

Fig. 1.6.2.2i. Phylogeny of some breeds of goldfish (Gene 430, 5, 2009).

Another closely related but distinct species, common carp Cyprinus carpio, was

also domesticated in China for ornamental purposes, begining from the Jin dynasty (265

420). Domesticated common carps are known as koi, and many varieties were bred in

China, Korea, and Japan (1.6.2.2j). Koi are as diverse as goldfishes in their coloration,

but not in their morphology.

48

Fig. 1.6.2.2j. Wild common carp Cyprinus carpio and its domesticated descendants,

varieties of koi.

A spectacular example of a very recent diversification of breeds of a fish is

provided by guppy Poecilia reticulata, a sexually dymorphic livebearing fish which was

domesticated only in the early XX century (Fig. 1.6.2.2e). Clearly, artificial selection can

lead to a very rapid origin of novel phenotypes, in the context of both anagenesis and

cladogenesis.

49

Fig. 1.6.2.2e. Guppy Poecilia reticulata. (top) Wildtype male (top) and female (bottom).

(bottom) Males from six domestic varieties (Nature Reviews Genetics 2, 207, 2001;

http://en.wikipedia.org/wiki/File:Guppy_breeds.jpg).

1.6.2.4. Cost of domestication

As far as the traits that are of utility to humans are concerned, domesticated

species are superior over their wild progenitors, often by a wide margin. However, in

many other traits domesticated species are suboptimal. This suboptimality may be

50

conditional. Domesticated varieties are not adapted to living outside human care, and

establishment of secondarily wild, feral populations originated from individuals which

escaped domestication is always accompanied by reversal evolution which restores many

aspects of primitive phenotypes (Fig. 1.1.1.7a; Fig. 1.6.2.2c (dingo); Fig. 1.6.2.4a). A

likely explanation of this phenomenon is trade-offs between adaptation to different

demands of selection: trait states favored by artificial selection and natural selection in

captivity may confer low fitness under natural selection in the wild, and vice versa. In

particular, artificial selection imposed for the only purpose of creating a monstrosity (Fig.

1.6.2.2h) leads to phenotypes that are optimal only under such selection, but are

suboptimal under any kind of natural selection. Also, if a particular adaptation is

beneficial in the wild, but not in captivity, long evolution under domestication can

destroy it simply because the mutations that do so accumulate without opposition from

natural selection (Fig. 1.5.3.5c).

Fig. 1.6.2.4a. Feral animals. (left) Mustangs, feral horses in North America. Native

horses went extinct in the New World at the beginning of the Holocene ~12,000, and

extant mustangs originated from domestic horses imported by the Europeans, starting

from 1493. (right) A feral pig from Hawaii. Feral populations of domestic pigs were

established in many areas, including North and South America, Australia, and Pacific

islands.

However, sequence-level data indicate that domesticated species also supper from

unconditional suboptimality, due to to accumulation of unconditionally deleterious

alleles. Such alleles confer low fitness under any environment, and under any feasible

51

mode of artificial selection, except, perhaps, selection on low motility or sexual activity

(see Kaidanov's data). In dogs, the ratio of the numbers of nonsynonymous vs.

synonymous allele replacements is ~50% higher in the course of evolution after

domestication than in wild wolves. The relative excess of the number of nonsynonymous

replacements was even higher in the course of evolution of domesticated rice. It is very

unlikely that most of these extra nonsynonymous replacements occurred under positive

selection, and a probably cause of this pattern is relaxation of negative selection which

made it possible for many mildly deleterious protein-altering mutations to reach fixation.

Indeed, among those nonsynonymous replacements that occurred in the course of

domestication, the proportion of those that radically alter the chemical properties of an

amino acid (e. g., a Cys > Tyr) repalcement is elevated, and such radical replacements are

unlikely to be adaptive. Apparently, genomes of species that were domesticated ~10,000

years ago have already accumulated thousands of mildly deleterious alleles (Trends in

Genetics 22, 126, 2006; Molecular Biology and Evolution 25, 2331, 2008).

According to the theory of Microevolution, there are at least three mechanisms

that can cause accumulation of midly but unconditionally deleterious alleles in a

domesticated lineage:

1) Reduced efficiency of negative selection due to low effective population size

(Section 2.2.5),

1) Side-effects of strong artificial selection, due to hitch-hiking of deleterious

alleles (Section 2.3.1),

2) Relaxation of negative natural selection due to strong artificial selection.

It is not yet clear which of these mechanisms is more important for the observed

accumulation of unconditionally deleterious alleles in domesticated species. However, it

is already obvious that extant populations of wlid progenitor species are an invaluable

resource for improvement of domesticated species. Although lacking some specific

desirable qualities, produced by artificial selection, wild progenitors have generally

healthier genotypes and may also possess specific useful alleles which were accidentally

lost in domesticated varieties.

Section 1.6.3. Evolution in experiments

52

Evolution can be studied in experiments which, of course, must be much longer

that a generation time of the experimental organism. Some experiments with

microorganisms included tens of thousands of generations of evolution, while the longest

experiments performed so far on macroorganisms included only hundreds of generations.

An experiment may consists of passive monitoring of evolution by natural selection in a

captive population living under a carefully controlled environment. Such evolution may

involve numerous adaptive allele replacements, often with substantial homoplasy, and

may produce lineages with rather different adaptations, capable of indefinite coexistence.

Alternatively, an experimental population may be subject to artificial selection or to

evolution with all forms of selection artificially relaxed. Almost any variable population

initially responds to artificial selection, although the rate of this response usually

decreases with time. Relaxed selection leads to rapid decline of fitness, measured under

harsh conditions, due to unopposed accumulation of deleterious mutations.

1.6.3.1. Evolution of captive populations

Domestication and subsequent artificial selection may be viewed as a series of

gigantic-scale inadvertent evolutionary experiments. However, evolutionary experiments

can also be performed deliberately. In the simplest case, researchers simply monitor

evolution of captive populations, without manipulating it in any way, except through

control over the environment. Such experiments are probably similar to how

domesticated plants evolved until recent times.

The most extensive evolutionary experiment of this kind is being performed on

Escherichia loci since 1988 in Michigan State University in the laboratory of Richard

Lenski. The experiment consists of 12 populations, initiated from the same ancestral

clone. Each population lives of 10 ml of minimal medium with glucose and is transferred

daily to fresh medium. Samples of bacteria were regularly stored at -80C, and used later

for sequencing and analysis of phenotypes. More than 6,000 days, and over 50,000

generations, lapsed so far since the beginning of the experiment, and a number of

fascinating evolutionary events have been observed during this period of time.

The first group of observations are concerned with anagenesis (compare to

Section 1.5.2.2). The environment of experimental populations is quite different from

usual environments of E. coli, which led to rapid adaptive evolution in each of the 12

53

populations. Comparison of genotypes from different generations revealed a large

number of allele replacements. Fig. 1.6.3.1a presents data on allele replacements in one

of the 2 populations. If a new allele appears in a genotype sequenced from some

generation of the experiment, and is also observed in genotypes sequenced from all

subsequent generations, we know that an allele replacement occurred.

Fig. 1.6.3.1a. Allele replacements detected by sequencing genomes sampled between

2,000 and 20,000 generations from one evolving population of E. coli. The outermost

ring represents the genome sampled at 20,000 generations, and labels all genes with SNP

mutations in coding (black) and intergenic (purple) regions, and those with DIP

mutations including deletions (red), insertions (green), insertion sequence (IS) element

insertions (blue), and an inversion between citC and gatZ (orange). Insertion sequences

54

are transposable elements present in bacterial genomes. The next four rings, from outer to

inner, show mutations present in genomes sampled at 15,000, 10,000, 5,000, and 2,000

generations. The innermost circle shows the genome position and scale in megabase pairs

(Mb). Mutations that are off the line of descent to a genome sampled at 40,000

generations are capped with a circle. Only one mutation (kup/insJ-5), a 1-base-pair (bp)

insertion near an IS150 element, shows an aberrant homoplastic distribution, being

present in clones 10K and 20K but not 15K (Nature 461, 1243, 2009).

A majority of allele replacements that occurred in the first 20,000 generations

were single-nucleotide substitutions, mostly within protein-coding regions which

comprise 85% of the E. coli genome. However, mutations of other kinds, including one

very long inversion, were also being fixed. Strikingly, all 26 substitutions that were fixed

during this time within protein-coding regions were missense substitutions, and no

synonymous or nonsense substitutions were observed. This strongly suggests that these

allele replacements were driven by substantial positive selection, which almost never

operates on synonymous mutations.

Thus, although the environment remained constant in the course of the

experiment, adaptive evolution due to positive selection-driven allele replacements is still

ongoing. Moreover, the rate of accumulation of allele replacements remained constant

until generation 26,500 and increased after that time, because the experimental

population evolved a higher mutation rate (Fig. 1.6.3.1b). However, the rate of fitness

increase declined, indicating that beneficial mutation that we fixed early conferred larger

selective advantages, perhaps because the population was further away from the fitness

peak. Still, it is too early to conclude when, and if, the mean population fitness will stop

growing.

55

Fig. 1.6.3.1b. Dynamics of allele replacements and of fitness improvement. Blue circles

show the total number of fixed mutations, relative to the ancestor, in each sampled clone.

The blue line represents a model where mutations accumulate uniformly over time. The

light blue curves define the 95% confidence interval for this linear model. Green squares

show the improvement of this populations mean fitness relative to the ancestor over

time, and the green curve is a hyperbolic plus linear fit of this trajectory. The inset shows

the number of mutations in the 40,000-generation clone; the dashed curve approximates

the change in the timecourse of genomic evolution after a mutator phenotype appeared by

about generation 26,500 (Nature 461, 1243, 2009).

The second group of observations on 12 evolving populations of E. coli are

concerned with patterns in independent anageneses (compare to Section 1.5.2.4). Twelve

instances of independent adaptation of the same ancestral genotype to the same

environment resulted in a strong parallelismat the level of the genes involved, but not of

particular mutations fixed. Two genes, nadR which encodes a transcriptional regulator,

and pykF which encodes pyruvate kinase, accepted an adaptive allele replacement in all

12 populations, although all these replacements were different in different populations.

Data on allele replacements in 4 genes are presented in Fig. 1.6.3.1c. It is not clear why

selection favored mutations in these genes.

56

Fig. 1.6.3.1c. Allele replacements that occurred by 20,000 generations in four genes in 12

experimental populations, designated -1, ..., -6 and +1, ..., +6. Lighter regions indicate

protein-coding sequences for and near pykF (A), nadR (B), pbpA-rodA (C), and

hokB/sokB (D). Each arrow marks the site of a replacement; the number shows the

affected population. The mutations in and near ydcA are of unknown relevance. All

mutations accepted in the coding regions, are nonsynonymous substitutions, except one

synonymous substitution and a 1-nucleotide insertion in ydcA in population 2.

Mutations fixed by 20,000 generations in 12 experimental populations of E. coli (PNAS

103, 9107, 2006).

Moreover, different populations independently evolved reduced catabolitic

function (Fig. 1.6.3.1d). This is not surprising, because glucose was the only source of

carbon and this function was not used. This evolution was faster in populations that

evolved elevated mutation rates than in populations that kept low ancestral mutation rate.

However, the difference was only marginal. Thus, direct selection for reduction of a

costly unused function, apparently played a substantial role in its reduction. Also, cells in

all populations became larger, and are now dividing substantially faster that at the

beginning of the experiment.

57

Fig. 1.6.3.1d. Evolution of total catabolic function.Values are shown on a log-

transformed scale, where the decay predicted by mutation accumulation is linear. Each

point is the mean of three clones from each of the 12 population. Solid line, mean value

of populations that retained the ancestral mutation rate at that generation; dashed line,

mean of three mutator populations; closed circles, populations with the ancestral

mutation rate; open circles, mutator populations; open triangle, late appearing mutator

(Nature 407, 736, 2000).

The third group of observations are concerned with cladogenesis. Several

populations at some point split into two or more lineages which coexisted later on,

indicating that they evolved different adaptations and occupied different ecological

niches (Section 2.1.1). In one populations, a cladogenesis occurred by generation 6000,

and the two lineages coexist thereafter. Individuals from these two lineages differ from

each other not only ecologically but also morphologically, and are designated L (large)

and S (small), accroding to the size of colonies they form on solid medium. When the

density of bacteria is low, L lineage grows exponentially about 20% faster than S lineage.

Still, two factors enable the stable coexistence of L and S lineages. First, L excretes a

metabolite that differentially promotes the growth of S. Second, L experiences increased

death during the stationary phase, after glucose is depleted, when S is present.

58

Through time, the frequency of S lineage was rising and falling several times

between about 1020% and 5090% of the total population. Also, each lineage

independently underwent periods of sharp decline in genetic variation, indicating

selective sweeps due to fixations of beneficial mutations within the lineage (Fig.

1.6.3.1e). Thus, both L and S lineages continued to adapt following their origin, and their

adaptive evolution contributed to fluctuations in relative abundance. However, better

adaptation of one lineage did not lead to extinction of the other lineage. In other words,

allele replacements were confined to individual lineages, and benefitial mutation which

arose within, say, an L individual eventually displaced all other L genotypes, but not S

genotypes. This makes S and L distinct, sympatric asexual populations, early on the path

to unlimited divergence.

59

Fig. 1.6.3.1e. (top) Trajectories of genetic diversity within the L lineage (circles and solid

line) and L lineage (squares and dashed line). Significant declines in genetic variation

between consecutive samples are indicated by asterisks. (middle) An adaptive allele

replacement leads to deep reduction of genetic variation, because the descendants of only

one individual take over the population. This phenomenon is called a selective sweep

(Chapter 2.4). Variation is them slowly restored by mutation. (bottom) Graphical

illustration of independent adaptive replacements in L and S lineages that begun soon

after their formation (J. of Molecular Evolution 61, 171, 2005).

Long-term ongoing evolution has also been observed on other systems. For

example, when bacteriophage fX174 evolved on a continuous supply of sensitive host

bacteria for 180 days (~13,000 phage generations), allele replacements accumulated in a

clock-like manner throughout the study, except for a low rate during the first 20 days, at a

rate of 1 replacement every 2 days. Several lines of evidence suggest that most of the

replacements were adaptive, including many of those that occurred at silent sites. Often,

several bacteriophages coinfect the same bacterium, and ongoing adaptive evolution may

be due to a potentially indefinite arms race within the bacteriophage population, due to

their competition within the same host cell.

60

Changes in the bacteriophage fX174 population. Solid diamonds show the number of

differences between each sequenced isolate and the ancestral sequence. Open triangles

show the distance between pairs of isolates from the same time point (Genetics 170, 19,

2005).

1.6.3.2. Long-term artificial selection

Let us now consider evolutionary experiments in which a population is not just

passively monitored but undergoes active artificial selection. Usually, an attempt to

change a phenotype by artificial selection is successful. After not too many generations

of selection, the mean value of a quantitative trait often moves outside the range of

variation of the trait within the intial population. However, response to artificial selection

eventually declines, due to the population running out of genetic variation and/or to

natural selection against phenotype that are favored by artificial selection.

Remarkably, the first crucial result obtained in an evolutionary experiment which

employed artificial selection was negative, in the sense that no response to selection has

been obtained. This experiment was performed by Wilhelm Johannsen, and the data were

published in 1903 (Journal of the History of Biology 7, 5, 1974). Johannsen worked with

beans Phaseolus vulgaris, which is an almost obligate selfer. Each plant is homozygous

and, thus, the genotype of an offspring is identical to the genotype of its only parent, as it

61

is also the case with apomixis (Section 2.3.3). Johannsen established several populations,

each originated from one parent and, thus, consisting of genetically identical plants

("pure line"), and performed artificial selection for low and high weight of seeds in the

course of a few generations. Within a pure line selection turned out to be completely

inefficient: the average weight of seeds produced by an offspring did not depend on the

weight of a seed from which it was grown. This negative result, caused by the lack of

heritable variation within a pure line, led Johannsen to formulate the concepts of

genotype and phenotype. Of course, if his experiment lasted longer and involved a larger

number of individuals, new mutations would generate enough genetic variation to allow

even a "pure line" to respond to selection. Indeed, populations of E. coli, described in the

previous Section (Fig. 1.6.3.1a), were also initiated as identical "pure lines", and if

Lenski's experiment lasted only for a few generations (instead of 50,000) and involved

only thousands (instead of billions of billions) of individuals no changes would occur in

his populations.

In fact, no astronomical numbers of individuals are needed for new mutations to

produce enough genetic variation to make a substantial response to selection possible.

Popular model traits often used for evolutionary experiments are the numbers of

abdominal and sternopleural bristles on Drosophila (Fig. 1.6.3.2a). In one experiment,

divergent artificial selection for abdominal and sternopleural bristle numbers was

performed in six populations, each of which originated from the same was highly inbred

and essentially genetically homogeneous initial population. Nevertheless, after 125

generations of selection an average divergence of 12.0 abdominal and 8.2 sternopleural

bristles was observed (Fig. 1.6.3.2b). Responce to selection decelerated with time, but

reverse selection always produced rapid responces, indicating that new mutations

affecting bristle number, which were selected for artificially, mostly have deleterious

effects on fitness. In fact declining fitness made it necessary to reduce the intensity of

artificial selection in the course of the experiment: before the 64th generations, only 25%

of flies with the highest or lowest bristle numbers were selection, but after that time is

became necessary to select 50% of flies, due to increased difficulty in obtaining enough

individuals.

62

Fig. 1.6.3.2a. Numbers of abdominal and sternopleural bristles in Drosophila are two

convenient and seemingly unimportant traits for artificial selection experiments. Here, D.

melanogaster is shown.

63

Fig. 1.6.3.2b. Responses to 125 generations of selection for abdominal (top) and

sternopleural (bottom) bristle number from an inbred base population. Closed symbols

and solid lines show generation means of high bristle number replicates, and open

symbols and dashed lines show generation means of low bristle number replicates.

Circles, triangles and squares indicate replicates 1, 2 and 3, respectively (Genetics 136,

937, 1994; Integrative and Comparative Biology 45, 391, 2005).

The longest continuous evolutionary experiment is known as The Illinois Long-

Term Selection Experiment. This experiment was initiated from an outbred, genetically

variable population and uses artificial selection for grain protein and oil concentration in

corn. Each generation 60 - 120 plants were scored, and seeds from the highest or lowest

20% of the distribution (depending on the direction of selection) were selected to form

the next generation. So far, a total of 103 generations of selection produced nine

populations that exhibit phenotypic extremes for grain composition and a host of

correlated traits (Fig. 1.6.3.2c). Remarkably, the significant amount of genetic variation

remaining in these populations. Indeed, selection response continues in all the

64

populations except ILO and ILP, which have probably reached lower biological limits for

these traits owing to their poor germination frequencies and lack of change in recent

cycles. However, reverse selection is invariably successful.

Fig. 1.6.3.2c. (a) Selection responses in the Illinois Protein Strains. Selection for high and

low concentration of protein in the grains has been performed for 103 generations and

produced the Illinois High Protein (IHP) and Illinois Low Protein (ILP) strains. Selection

65

was reversed in these strains beginning at generation 48 to produce the Reverse High

Protein (RHP) and Reverse Low Protein (RLP) strains. Reverse Low Protein 2 (RLP2)

strain was initiated